Abstract

To investigate the characteristics of anaerobic ammonia oxidation for treating low-ammonium wastewater, a continuous-flow completely autotrophic nitrogen removal over nitrite (CANON) biofilm reactor was studied. At a temperature of 32 ± 1 °C and a pH between 7.5 and 8.2, two operational experiments were performed: the first one fixed the hydraulic retention time (HRT) at 10 h and gradually reduced the influent ammonium concentrations from 210 to 50 mg L−1; the second one fixed the influent ammonium concentration at 30 mg L−1 and gradually decreased the HRT from 10 to 3 h. The results revealed that the total nitrogen removal efficiency exceeded 80%, with a corresponding total nitrogen removal rate of 0.26 ± 0.01 kg N m−3 d−1 at the final low ammonium concentration of 30 mg L−1. Small amounts of nitrous oxide (N2O) up to 0.015 ± 0.004 kg m−3 d−1 at the ammonium concentration of 210 mg L−1 were produced in the CANON process and decreased with the decrease in the influent ammonium loads. High-throughput pyrosequencing analysis indicated that the dominant functional bacteria ‘Candidatus Kuenenia’ under high influent ammonium levels were gradually succeeded by Armatimonadetes_gp5 under low influent ammonium levels.

INTRODUCTION

Anaerobic ammonia oxidation (anammox) plays a key role in the removal of biological nitrogen from ammonium wastewater. To remove ammonium-nitrogen, the completely autotrophic nitrogen removal over nitrite (CANON) process, a typical and popular anammox technique, depends on the harmony and equilibrium of the interactions between aerobic ammonia-oxidizing bacteria (AerAOB) and anammox bacteria (AnAOB) in a single-stage reactor under oxygen-limited conditions (Sliekers et al. 2002). First, NH4+ is oxidized to NO2 by the AerAOB, following which the remaining NH4+ and the generated NO2 are converted to N2 by the AnAOB (Liang et al. 2014). Compared to traditional nitrification and denitrification processes, the CANON process has numerous advantages, including reduced aeration without additional organic carbon, less sludge production, reduced operational and construction costs, and enhanced sustainability and economic benefits (Zhang et al. 2012).

At present, CANON technology has been successfully applied in various industrial wastewater treatments, such as sludge digestion, landfill leachate and other high-ammonium-nitrogen, low-carbon wastewaters (Zhang et al. 2013a). Nevertheless, studies on the use of CANON technology to treat low-ammonium wastewater are still in the exploration stage, and several systems have been operated using lower ammonium concentrations (≤100 mg L−1), such as domestic wastewater and secondary effluent sewage. One previous report proved that a CANON reactor initiated with an influent ammonium concentration of 400 mg L−1 achieved an excellent performance for treating synthetic wastewater containing 100 mg L−1 ammonium (Liu et al. 2012a). Another study also confirmed that it was feasible to start up a CANON to treat domestic sewage in a membrane bioreactor (MBR) with a low ammonium concentration (Zhang et al. 2013a). Furthermore, a previous study also concluded that the nitrogen removal rate achieved a value of 0.26 kg N m−3 d−1 in a gaslift reactor with anammox biomass when the temperature and influent ammonium concentration were 20 °C and 69 mg L−1, respectively (Hendrickx et al. 2012). However, very little information regarding the use of the CANON process to treat different types of low-ammonium wastewater is available. Thus, this study is relevant to the application of the CANON process to various low-ammonium wastewaters.

Nitrous oxide (N2O), as a powerful greenhouse gas in the atmosphere, is capable of seriously damaging the ozone layer, in addition to having a global warming potential of approximately 310 times that of carbon dioxide (IPCC 2007). At present, the consensus holds that AnAOB do not possess the pathway to produce N2O. Although no decisive conclusion has been reached, the phenomenon associated with the production of a small amount of N2O when using AnAOB has been studied in some previous investigations (Lotti et al. 2014; Harris et al. 2015). Small amounts of N2O (on average, accounting for an influent total nitrogen (TN) load of 2.7‰) are detected in anammox reactors fed with material with high ammonium concentrations (Kampschreur et al. 2009; Okabe et al. 2011); therefore, the anammox process is certainly not a major pathway of N2O production (Okabe et al. 2011). One study confirmed that the hydroxylamine oxidation pathways contributed most of the N2O emissions via the coupling of nitritation and anammox processes in partial nitrification reactors that are supplied with high-ammonium synthetic wastewater (Rathnayake et al. 2013). In a single-stage autotrophic nitrogen removal reactor fed with an NH4+ concentration of 300 mg L−1, the processes of NH2OH oxidation and nitrifier denitrification are likely responsible for the majority of the total N2O production in the aerobic area, and the remaining N2O produced in the AnAOB-dominated anoxic area is likely generated via heterotrophic denitrification (Muhammad et al. 2016). Moreover, heterotrophic denitrification may be the major pathway of N2O production in the anoxic zone in a sequencing batch reactor fed with an NH4+ concentration of 816 ± 13 mg L−1 (Wang et al. 2014). Nevertheless, information on N2O production remains relatively rare for the CANON process with respect to the treatment of low-ammonium wastewater; thus, N2O production shall also be investigated for low influent ammonium conditions.

During the past decades, five genera of anammox bacteria, including 13 species, have been identified. Influent NH4+-N is always regarded as an important substrate for the CANON process in the treatment of wastewater because it affects the living environments of AerAOB and AnAOB, thereby giving rise to changes in the community structure (Liang et al. 2014). In a CANON biofilter, the microbial characteristics have been found to change with decreasing ammonium concentrations (Liu et al. 2012a). Several previous studies have indicated that Nitrosomonas (AerAOB) are the dominant bacteria in CANON reactors (Ballinger et al. 1998; Zhang et al. 2013a). Another study reached the conclusion that ‘Candidatus Kuenenia stuttgartiensis’ existed in a CANON reactor fed with a high ammonium concentration (Liang et al. 2014). Similarly, ‘Candidatus Kuenenia stuttgartiensis’ was also identified as the dominant functional bacteria in the CANON process in association with influent wastewaters with ammonium concentrations of 256–666 mg L−1 (Vázquez-Padín et al. 2010). Consequently, it is valuable for bacterial engineering work to obtain information on the dominant functional bacteria in a CANON reactor. Nevertheless, for the CANON process, few specific studies have been performed to detect the changes in the dominant functional bacteria with decreasing influent ammonium concentrations. Thus, it is very valuable for us to study the dominant functional bacteria in low-ammonium wastewater.

In this study, the CANON process in a continuous-flow biofilm reactor was conducted to investigate the characteristics of the anammox process in the treatment of low-ammonium wastewater. The treatment performance, nitrous oxide production and the microbial community associated with low-ammonium wastewater were studied. This will provide a valuable reference for future treatment of low-ammonium wastewater.

MATERIALS AND METHODS

Experimental facility

A continuous-flow biofilm reactor was constructed for this experiment. Details of the reactor are shown in Figure 1. It was a rectangular column made of plexiglass (18.3 cm in length, 18.3 cm in width and 37.5 cm in height). The outer wall of this reactor was surrounded by opaque plastic for protection from light exposure. The fillers in the reactor were plastic fibers. The effective working volume of the reactor was 7 L. Because the inlet of the silicone tube was located at the bottom of the reactor, influent was continuously introduced to the bottom of the tank, and discharge was released from the upper outlet. Additionally, a peristaltic pump in the reactor ensured that all substances in the water were mixed completely while avoiding large disturbances to the biofilm.

Figure 1

Experimental facility and process schematic chart.

Figure 1

Experimental facility and process schematic chart.

Seeding sludge and feeding media

The activated sludge inoculated in this reactor was extracted from the wastewater treatment plant of the Chang'an District in Xi'an, Shaanxi, China. The synthetic wastewater was a mixture of the appropriate amounts of NH4HCO3, NaHCO3, anhydrous CaCl2, KH2PO4, MgSO4•7H2O and a trace element solution (Lovley & Phillips 1998) in tap water. Influent NH4+-N concentration was gradually decreased from 210 to 136, 100, 70, 50 and 30 mg L−1 by reducing the quantity of NH4HCO3. The pH was maintained at a relatively stable value by changing the quantity of NaHCO3. The concentrations of other ingredients in the synthetic water were constant: MgSO4•7H2O 11.07 mg L−1, KH2PO4 21.67 mg L−1, anhydrous CaCl2 11.07 mg L−1 and trace element solution 1 mL per 7.5 L. In addition, the tap water contained very little NH4+, chemical oxygen demand, NO2 and NO3. Additionally, there was no need to drive out the oxygen in the synthetic wastewater.

Experimental background and operation strategy

The original CANON reactor stably treated wastewater with higher influent ammonium-nitrogen concentrations for approximately 5 months and successfully achieved an excellent performance, with a TN removal efficiency and a corresponding total nitrogen removal rate (TNRR) of 80% and 1.02 kg N m−3 d−1, respectively. This experiment represents a study based on the original CANON reactor that aims to investigate the performance of the CANON reactor by gradually changing the influent ammonium load.

The temperature was held at 32 ± 1 °C by a thermostatic heater in the water, the pH value varied between 7.5 and 8.2 (mean of 8.0), and the dissolved oxygen (DO) was held at no more than 1.2 mg L−1 through adjustments to the aeration rate. Two operational stages were used to change the TN load. The first stage involved a constant hydraulic retention time (HRT) of 10 h and a gradual decrease in the influent ammonium concentration (with values of 210, 136, 100, 70 and 50 mg L−1). The second stage involved a constant influent ammonium concentration of 30 mg L−1 and a gradual decrease in the HRT (with values of 7, 5 and 3 h).

Analytical methods

In accordance with the standard methods (Chinese SEPA 2002), concentrations of NH4+-N, NO2-N and NO3-N were measured daily using Nessler's reagent spectrophotometry, the N-(1-naphthyl)-ethylenediamine photometric method and UV spectrophotometry, respectively. The DO, temperature and pH were measured with a Hach-HQ30d DO meter, a TC-058 intelligent temperature measurement device and an S10 portable intelligent pH meter, respectively. TN was calculated based on the sum total of NH4+-N, NO3-N and NO2-N.

The dissolved N2O was monitored by a N2O microsensor (Unisense, Denmark). Dissolved N2O in the CANON reactor would also have been emitted to the air by stirring and air stripping. A previous study has shown that N2O production can be estimated using a linear equation based on the soluble N2O (Zhao et al. 2016). N2O production in this continuous-flow reactor was calculated using the following Equation (1):
formula
(1)
where was the total N2O production rate (mg L−1 min−1), and Equation (1) was a combination of two equations: and in this continuous-flow reactor; re was the N2O emission rate (mg L−1 min−1), rc was the N2O accumulation rate (mg L−1 min−1) was the dissolved N2O concentration (mg L−1), V was the effective volume of the CANON reactor, Q represented the flow rate in this experiment, and K was the N2O emission coefficient, which was determined through a series of special experiments as shown in Table 1. In a container of the same shape and volume of the reactor, the same volume of tap water was injected, and the same stirring rate and same aeration rate were maintained. Then, N2O gas with a purity of 99.9% was injected into the reactor. The liquid N2O concentrations were measured with a N2O microelectrode, and the release rate of N2O was obtained. The escape curve at different aeration rates was obtained by changing the aeration rate.
Table 1

N2O emission coefficient (K) under different aeration rates in this experiment

Aeration flow (mL min−1)K (min−1)
0.006 
20 0.006 
30 0.009 
50 0.012 
100 0.015 
150 0.018 
200 0.025 
240 0.036 
Aeration flow (mL min−1)K (min−1)
0.006 
20 0.006 
30 0.009 
50 0.012 
100 0.015 
150 0.018 
200 0.025 
240 0.036 

Microbial community analysis

Sludge samples extracted from the CANON reactor were used to identify the microbial community structures when the influent ammonium concentrations were 210 mg L−1 (the sludge sample was obtained when the ammonium-nitrogen concentration in influent was decreased from 500 to 210 mg L−1 and run for 30 days prior to the present experiment) and 30 mg L−1 (the 140th day in this experiment). Microbial communities were identified via the high-throughput 16 S rRNA gene sequencing technique. The DNA of the sludge sample was extracted using the OMEGA DNA kit for soil according to the instructions of the manufacturer. The pyrosequence data were analyzed using the 16 S Ribosomal Database project classifier, and similar sequences were assigned to the same operational taxonomic units (OTUs) based on a 97% similarity.

RESULTS AND DISCUSSION

CANON reactor operation

The influent nitrogen compound content, total nitrogen loading rate (TNLR), and the TNRR are summarized in Figure 2(a). The first stage fixed the HRT at 10 h, and the influent ammonium concentrations were gradually reduced from 210 to 136, 100, 70 and 50 mg L−1; the second stage fixed the influent ammonium concentration at 30 mg L−1 and the HRT was gradually reduced from 7 to 5 to 3 h. A clear positive correlation exists between the TNLR and TNRR in Figure 2(a). The effluent nitrogen-compound content, TN concentration, and the removal efficiencies of TN and ammonium-nitrogen are shown in Figure 2(b). The NO2-N concentration was low, and the concentrations of ammonium and nitrate fluctuated early on, which led to fluctuations in the removal efficiencies of NH4+-N and TN. The CANON reactor took approximately 4 to 7 days to reach a steady state when the TN loads were changed.

Figure 2

Effects of the CANON reactor on nitrogen removal: (a) the concentrations of various influent nitrogen compounds and the TNLR and TNRR; (b) the concentrations of various effluent nitrogen compounds and TN, and the TN and NH4+-N removal efficiencies.

Figure 2

Effects of the CANON reactor on nitrogen removal: (a) the concentrations of various influent nitrogen compounds and the TNLR and TNRR; (b) the concentrations of various effluent nitrogen compounds and TN, and the TN and NH4+-N removal efficiencies.

The average TNLR and TNRR levels are shown in Table 2. The data in Table 2 were calculated based on the relatively stable and good removal performance stage of the CANON process. Consequently, the optimal TNRR was achieved at 0.26 ± 0.01 kg N m−3 d−1 when this CANON reactor was fed with an ammonium concentration of 30 mg L−1. In previous studies, TNRR values of 0.26, 0.09 and 0.25 kg N m−3 d−1 were achieved when the influent ammonium concentrations were 69, 60, and 116.7 mg L−1, respectively (Vázquez-Padίn et al. 2009; Hendrickx et al. 2012; Lotti et al. 2014). This study also demonstrated that the CANON process achieved an excellent performance. More precisely, the removal efficiency of the CANON process for each influent ammonium concentration stabilized at 90% or more, and the effluent NH4+-N was almost zero. The TN removal efficiency also stabilized at approximately 80% under various TN loads, indicating that the CANON process can adapt well to nitrogen load variations and can efficiently remove nitrogen from synthetic wastewater. Several previous studies found maximum TN removal efficiencies of approximately 88.8%, 85%, 89% and 80% (Vázquez-Padίn et al. 2009; Hendrickx et al. 2012; Liu et al. 2012a; Zhang et al. 2013a, respectively). The putative explanation for this phenomenon can involve a theoretical limit on the maximum TN removal efficiency (Vázquez-Padίn et al. 2009) in the CANON process and a mass transfer limitation in the biofilm, which requires certain concentrations of ammonium-nitrogen and nitrite-nitrogen.

Table 2

Summary of the changes in water quality in the CANON process

StageInfluent NH4+ (mg L−1)TNLR (kg N m−3 d−1)TNRR (kg N m−3 d−1)ΔNH4+ (mg L−1)ΔNO3 (mg L−1)TN removal efficiency (%)
Stage I 210 0.57 ± 0.02 0.39 ± 0.04 187.97 ± 22.45 33.22 ± 11.27 81.03 
136 0.38 ± 0.03 0.30 ± 0.04 119.87 ± 18.34 16.27 ± 4.94 85.27 
100 0.26 ± 0.01 0.20 ± 0.02 80.94 ± 12.09 10.79 ± 3.16 83.88 
70 0.19 ± 0.01 0.14 ± 0.02 53.16 ± 7.98 8.10 ± 2.62 84.91 
50 0.13 ± 0.01 0.11 ± 0.06 44.50 ± 2.70 6.12 ± 0.92 85.72 
Stage II 30 0.11 ± 0.01 0.09 ± 0.01 28.14 ± 1.10 4.16 ± 1.30 84.65 
30 0.16 ± 0.01 0.12 ± 0.01 26.54 ± 2.21 4.33 ± 1.73 86.93 
30 0.32 ± 0.01 0.26 ± 0.01 29.04 ± 0.90 3.08 ± 0.61 84.48 
StageInfluent NH4+ (mg L−1)TNLR (kg N m−3 d−1)TNRR (kg N m−3 d−1)ΔNH4+ (mg L−1)ΔNO3 (mg L−1)TN removal efficiency (%)
Stage I 210 0.57 ± 0.02 0.39 ± 0.04 187.97 ± 22.45 33.22 ± 11.27 81.03 
136 0.38 ± 0.03 0.30 ± 0.04 119.87 ± 18.34 16.27 ± 4.94 85.27 
100 0.26 ± 0.01 0.20 ± 0.02 80.94 ± 12.09 10.79 ± 3.16 83.88 
70 0.19 ± 0.01 0.14 ± 0.02 53.16 ± 7.98 8.10 ± 2.62 84.91 
50 0.13 ± 0.01 0.11 ± 0.06 44.50 ± 2.70 6.12 ± 0.92 85.72 
Stage II 30 0.11 ± 0.01 0.09 ± 0.01 28.14 ± 1.10 4.16 ± 1.30 84.65 
30 0.16 ± 0.01 0.12 ± 0.01 26.54 ± 2.21 4.33 ± 1.73 86.93 
30 0.32 ± 0.01 0.26 ± 0.01 29.04 ± 0.90 3.08 ± 0.61 84.48 

Note: ΔNH4+ represents the reduction in the influent ammonium concentration; ΔNO3 represents the production of nitrate concentration.

The production of nitrate (ΔNO3) and the reduction of NH4+ (ΔNH4+) under each TNLR are presented in Table 2. The molar concentration ratios of ΔNO3 to ΔNH4+ were 0.16 ± 0.05:1, 0.13 ± 0.06:1, 0.11 ± 0.03:1, 0.14 ± 0.03:1 and 0.13 ± 0.02:1 in stage I and 0.14 ± 0.04:1, 0.15 ± 0.06:1 and 0.11 ± 0.02:1 in stage II. As previously identified, the stoichiometric equation for the CANON process is expressed by the following equation (Third et al. 2001):
formula
(2)

According to Equation (2), the theoretical value of ΔNO3 to ΔNH4+ is 0.13:1. The errors of this test result were +23% and −15% of the theoretical value because of the fluctuation of DO. Moreover, a previous study also reported that the actual ratio of ΔNO3/ΔNH4+ was less than the theoretical ratio obtained in the CANON process as a consequence of the accumulation of nitrite in the CANON reactor (Zhang et al. 2013b).

Effect of DO on nitrogen removal in CANON reactor

The DO concentration is an important factor affecting the anammox reaction. To achieve the goal of retaining only ammonia-oxidizing bacteria (AOB) and gradually eliminating nitrite-oxidizing bacteria (NOB) and to finally obtain stable partial nitrification in the CANON reactor, a low DO level is required, as the affinity of AOB for DO exceeds that of NOB (Pathak et al. 2007). However, high DO concentrations could improve the efficiency of the partial nitrification process despite increasing the disadvantage of converting NO2-N to NO3-N, which is harmful to AnAOB activity. Therefore, an optimum DO level should be maintained to balance the interactions between AerAOB and AnAOB. For the different influent TNLRs, different DO levels were used. Based on the measurements of DO concentrations and TN removal efficiencies throughout this experiment, the various DO concentrations and corresponding TN removal efficiencies are shown in Figure 3.

Figure 3

Effects of DO on TN removal efficiency ((a), (b), (c), (d) and (e) represent the TNLR values of 0.57 ± 0.02, 0.38 ± 0.03, 0.26 ± 0.01, 0.19 ± 0.01 and 0.13 ± 0.01 kg N m−3 d−1, respectively, in stage I at an HRT of 10 h; (f), (g) and (h) represent the TNLR values of 0.11 ± 0.01, 0.16 ± 0.01 and 0.32 ± 0.01 kg N m−3 d−1, respectively, in stage II at a TN concentration of 30 mg L−1).

Figure 3

Effects of DO on TN removal efficiency ((a), (b), (c), (d) and (e) represent the TNLR values of 0.57 ± 0.02, 0.38 ± 0.03, 0.26 ± 0.01, 0.19 ± 0.01 and 0.13 ± 0.01 kg N m−3 d−1, respectively, in stage I at an HRT of 10 h; (f), (g) and (h) represent the TNLR values of 0.11 ± 0.01, 0.16 ± 0.01 and 0.32 ± 0.01 kg N m−3 d−1, respectively, in stage II at a TN concentration of 30 mg L−1).

The CANON reactor always had an optimum DO for nitrogen removal in both stage I (constant HRT and decreasing TNLR through decreasing the influent NH4+-N concentration) and stage II (constant influent NH4+-N concentration and decreasing the HRT to increase the TNLR). The TN removal efficiency was more than 80%, and the NH4+ removal efficiency was more than 90%. The optimum DO concentrations corresponding to Figure 3(a)3(h) were 0.8, 0.75, 0.5, 0.35, 0.28, 0.25, 0.36 and 0.63 mg L−1, respectively. In contrast, a previous study found that the optimum DO concentration for nitrogen removal via an activated sludge process in a CANON reactor supplied with an ammonium-nitrogen concentration of 116.7 mg L−1 was 0.5 mg L−1 (Vázquez-Padίn et al. 2009). Another study achieved a TNRR of 0.97 kg m−3 d−1 by maintaining a DO concentration of 0.15 mg L−1 during the treatment of domestic sewage in an MBR-CANON system (Zhang et al. 2013a). Additionally, another study indicated that the maximum TNRR values in a CANON biofilter were achieved when the influent ammonium-nitrogen concentrations of 100 and 200 mg L−1 had corresponding DO concentrations of 5.06 ± 0.86 and 6.72 ± 0.83 mg L−1, respectively (Liang et al. 2014). Differences in experimental parameters (such as the HRT, pH and temperature) and different treatment processes (such as using biofilm or an activated sludge process) between this study and previous studies could account for the large differences in the experimental results. However, the results demonstrate that there is always an optimum DO concentration corresponding to the different influent nitrogen loading rates which results in the maximum TNRR. Moreover, the optimum DO concentration decreases with decreasing TNLR. In addition, because the TNLR was changed by reducing the influent ammonium concentration in stage I and reducing the HRT in stage II, the optimum DO concentration likely decreased with the decrease in the influent ammonium-nitrogen concentration and increased with the reduction in the HRT.

However, the TNRR decreases when an improper aeration rate causes the DO concentration to exceed the optimum value in the CANON reactor. This process is due to the competition between NOB and AnAOB for NO2 and to the excessive conversion of NO2 to NO3. Eventually, excessive DO concentrations can lead to a decrease in the nitrogen removal efficiency in the CANON process.

N2O production

Because the N2O emission coefficient fluctuated under various aeration rates, the amount of N2O production was calculated using Equation (1) as presented in the ‘Analytical methods’ section. The dissolved N2O concentration, N2O emission rate, N2O production and the correlation between N2O production and TNLR are depicted in Figure 4.

Figure 4

N2O production: (a) the correlation between N2O production and TNLR (a, b, c, d, e, f, g and h represent different conditions: the TN concentrations of 210 (HRT = 10 h), 136 (HRT = 10 h), 100 (HRT = 10 h), 70 (HRT = 10 h), 50 (HRT = 10 h), 30 (HRT = 7 h), 30 (HRT = 5 h) and 30 mg L−1 (HRT = 3 h), respectively); (b) dissolved N2O concentration, N2O production and N2O emission rate.

Figure 4

N2O production: (a) the correlation between N2O production and TNLR (a, b, c, d, e, f, g and h represent different conditions: the TN concentrations of 210 (HRT = 10 h), 136 (HRT = 10 h), 100 (HRT = 10 h), 70 (HRT = 10 h), 50 (HRT = 10 h), 30 (HRT = 7 h), 30 (HRT = 5 h) and 30 mg L−1 (HRT = 3 h), respectively); (b) dissolved N2O concentration, N2O production and N2O emission rate.

The correlation between the N2O production and TNLR is shown in Figure 4(a). Under the conditions of a, b, c, d and e, the influent TNLR was reduced by decreasing the influent NH4+-N concentration, while the proportion of N2O production accounting for the influent TNLR (2.60 ± 0.52 to 4.49 ± 0.17%) rose. In contrast, the proportion of N2O production accounting for the influent TNLR (6.00 ± 0.10 to 3.85 ± 0.86%) decreased gradually when the ammonium-nitrogen concentration in influent was fixed at 30 mg L−1 under conditions of f, g and h, and the TNLR gradually increased with increasing influent flow. One explanation for these phenomena could be that the change in N2O production was much smaller than that in the TNLR. One previous study revealed that the proportion of N2O production accounting for the influent TNLR was 3.0–6.4% in a single-stage nitritation-anammox reactor fed with a TNLR of 0.58–2.7 kg N m−3 d−1 (De Clippeleir et al. 2013). Moreover, according to the calculation, the ratio of N2O production to the NH4+-N removed was 3.10 ± 0.87 to 7.50 ± 0.30% when this reactor was fed with a TNLR of 0.57 ± 0.02 to 0.11 ± 0.01 kg N m−3 d−1. In addition, the dissolved N2O concentration, N2O emission rate and N2O production appear to have the same tendencies in Figure 4(b). Under the conditions of a, b, c, d, e, f, g and h in Figure 4(b), when the influent ammonium load was high, the dissolved N2O concentration and N2O production were high, and the N2O emission rate was also high due to the greater aeration rate required. Based on Figure 4(a) and 4(b), the amount of N2O production was 28.8 ± 1.0 to 70.8 ± 20.0 μg min−1 and gradually decreased with the decrease in influent ammonium loads.

Generally, this process would produce a small amount of N2O, up to 0.015 ± 0.004 kg m−3 d−1, at the ammonium concentration of 210 mg L−1 in the CANON reactor, which has a minimally negative impact on the environment.

Microbial community structure in the CANON reactor

High-throughput 16S rRNA gene sequencing technology was used to identify the microbial communities at influent ammonium concentrations of 210 and 30 mg L−1 to investigate the changes in the microbial community in the CANON reactor. The results provided comprehensive and deeper insight into the microbial community compositions in the CANON reactor and could be very valuable for microbial engineering efforts. Because sequencing analysis suggested complex microbial diversity in the biofilm, the main bacteria at the level of phylum and genus are shown in Figure 5.

Figure 5

Pie charts showing the microbial community composition at the level of phylum and genus: (a) the microbial community associated with high influent ammonium concentrations; (b) the microbial community associated with low ammonium concentrations.

Figure 5

Pie charts showing the microbial community composition at the level of phylum and genus: (a) the microbial community associated with high influent ammonium concentrations; (b) the microbial community associated with low ammonium concentrations.

As shown in Figure 5(a), the phylum Planctomycetes was the most dominant phylum in the biofilm, accounting for 24.37%, when the influent ammonium concentration was 210 mg L−1. As shown in Figure 5(b), the phylum Proteobacteria was the most dominant phylum in this CANON reactor, accounting for 32.41%, when the influent ammonium concentration was 30 mg L−1. One study has revealed that one of the most dominant AerAOB in the CANON process was Proteobacteria (Liu et al. 2012b); thus, the large proportion of Proteobacteria present in this CANON reactor contributes to the partial nitrification process and the nitrite accumulation. The other dominant phyla were Proteobacteria, Armatimonadetes, Bacteroidetes, Chloroflexi and Firmicutes when the influent ammonium concentration was 210 mg L−1, as shown in Figure 5(a), and Planctomycetes, Armatimonadetes, Bacteroidetes, Chloroflexi and Nitrospirae when the influent ammonium concentration was 30 mg L−1, as shown in Figure 5(b). The proportions of Proteobacteria and Armatimonadetes in Figure 5(b) are much larger than those in Figure 5(a), which means Proteobacteria and Armatimonadetes readily adapted to the low-ammonium and oxygen-limited conditions and consequently promoted the anammox process in the treatment of low-ammonium wastewater in this CANON reactor.

As shown in Figure 5, the genus Nitrosomonas was not only the most important AerAOB genus under high NH4+-N concentrations, but also the most important AerAOB genus under low NH4+-N concentrations in this study. The AerAOB Nitrosomonas was previously identified as the dominant functional bacteria in a number of anammox reactors used for nitrogen removal (Ballinger et al. 1998; Zhang et al. 2013a). The proportion of Nitrosomonas (6.87%) in Figure 5(b) is much greater than that in Figure 5(a) (1.98%), which means Nitrosomonas played a more important role in low-ammonium environments than in high-ammonium environments.

The AnAOB genus ‘Candidatus Kuenenia’ (22.73%) was the most dominant functional bacteria in the high-ammonium environment. A previous study reported that ‘Candidatus Kuenenia’ was the main population in an anammox reactor fed with NH4+-N concentrations of 256–666 mg L−1 (Vázquez-Padín et al. 2010). Nevertheless, as the influent NH4+-N concentrations changed from high to low, the microbial community gradually altered. Ultimately, the AnAOB genus ‘Candidatus Anammoxoglobus’ (16.77%) became the dominant functional bacteria under low ammonium concentrations. Additionally, this study showed that ‘Candidatus Anammoxoglobus’ had the ability to adapt to the low ammonium concentration of 30 mg L−1 and can outcompete other AnAOB during nitrogen removal.

In addition, the genus Armatimonadetes_gp5 experienced a large change and increased from 4.29% under the high ammonium concentration to 20.45% under the low ammonium concentration. Furthermore, the genus Armatimonadetes_gp5 became the most dominant functional bacteria at the ammonium concentration of 30 mg L−1 and accounted for the largest proportion. Therefore, these bacteria are inferred to be a new type of AnAOB that promotes the anammox process during the treatment of low ammonium concentrations in the CANON reactor.

Based on Figure 5(a) and 5(b), the NOB Nitrospira (3.14%) appeared under the low ammonium concentration but not under the high ammonium concentration, which could be due to the small amount of free ammonia and inhibition mitigation under the low-ammonium conditions (Liu et al. 2012a).

CONCLUSIONS

  • In the CANON reactor, the influent ammonium concentrations were gradually reduced from 210 to 30 mg L−1. The TN removal efficiency reached 84.48% in the final stages, corresponding to a TNRR of 0.26 ± 0.01 kg N m−3 d−1.

  • The optimal DO concentrations decreased with decreasing influent ammonium loads and ranged from 0.25 to 0.8 mg L−1, corresponding to TNRR values of 0.09 to 0.47 kg N m−3 d−1, respectively.

  • The small amount of N2O produced by the CANON process accounted for 2.60 ± 0.52 to 6.00 ± 0.10% of the influent TNLR when this reactor was fed with a TNLR of 0.57 ± 0.02 to 0.11 ± 0.01 kg N m−3 d−1. In addition, the ratio of N2O production to the NH4+-N removed was 3.10 ± 0.87 to 7.50 ± 0.30% when this reactor was fed with a TNLR of 0.57 ± 0.02 to 0.11 ± 0.01 kg N m−3 d−1. The production of N2O reached 0.015 ± 0.004 kg m−3 d−1 at the ammonium concentration of 210 mg L−1 and decreased with decreasing influent ammonium loads.

  • The dominant functional bacteria ‘Candidatus Kuenenia’ under high influent ammonium conditions was gradually succeeded by Armatimonadetes_gp5 under low influent ammonium conditions.

ACKNOWLEDGEMENTS

This work was supported by the National Natural Science Foundation of China (Grant No. 51778057).

AUTHOR DISCLOSURE STATEMENT

No competing financial interests exist.

REFERENCES

Ballinger
,
S. J.
,
Head
,
I. M.
,
Curtis
,
T. P.
&
Godley
,
A. R.
1998
Molecular microbial ecology of nitrification in an activated sludge process treating refinery wastewater
.
Water Science and Technology
37
(
4–5
),
105
108
.
Chinese SEPA
2002
Water and Wastewater Monitoring Methods
(4th edn).
Chinese Environmental Science Publishing House
,
Beijing
,
China
.
De Clippeleir
,
H.
,
Vlaeminck
,
S. E.
,
Mosquera
,
M.
,
Boeckx
,
P.
,
Verstraete
,
W.
&
Boon
,
N.
2013
One-stage partial nitritation/anammox at 15°C on pretreated sewage: feasibility demonstration at lab-scale
.
Applied Microbiology and Biotechnology
97
(
23
),
10199
10210
.
Harris
,
E.
,
Joss
,
A.
,
Emmenegger
,
L.
,
Kipf
,
M.
,
Wolf
,
B.
,
Mohn
,
J.
&
Wunderlin
,
P.
2015
Isotopic evidence for nitrous oxide production pathways in a partial nitritation-anammox reactor
.
Water Research
83
,
258
270
.
Hendrickx
,
T. L. G.
,
Wang
,
Y.
,
Kampman
,
C.
,
Zeeman
,
G.
,
Temmink
,
H.
&
Buisman
,
C. J. N.
2012
Autotrophic nitrogen removal from low strength waste water at low temperature
.
Water Research
46
(
7
),
2187
2193
.
IPCC
2007
Climate Change 2007: the Physical Science Basis: Working Group I Contribution to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change
.
Cambridge University Press
,
Cambridge, UK
.
Kampschreur
,
M. J.
,
Temmink
,
H.
,
Kleerebezem
,
R.
,
Jetten
,
M. S. M.
&
van Loosdrecht
,
M. C. M.
2009
Nitrous oxide emission during wastewater treatment
.
Water Research
43
(
17
),
4093
4103
.
Liang
,
Y.
,
Li
,
D.
,
Zhang
,
X.
,
Zeng
,
H.
,
Yang
,
Z.
,
Cui
,
S.
&
Zhang
,
J.
2014
Nitrogen removal and microbial characteristics in CANON biofilters fed with different ammonia levels
.
Bioresource Technology
171
,
168
174
.
Liu
,
T.
,
Li
,
D.
,
Zeng
,
H.
,
Li
,
X.
,
Zeng
,
T.
,
Chang
,
X.
,
Cai
,
Y.
&
Zhang
,
J.
2012a
Biodiversity and quantification of functional bacteria in completely autotrophic nitrogen-removal over nitrite (CANON) process
.
Bioresource Technology
118
,
399
406
.
Liu
,
T.
,
Li
,
D.
,
Zeng
,
H.
,
Li
,
X.
,
Liang
,
Y.
,
Chang
,
X.
&
Zhang
,
J.
2012b
Distribution and genetic diversity of functional microorganisms in different CANON reactors
.
Bioresource Technology
123
,
574
580
.
Lotti
,
T.
,
Kleerebezem
,
R.
,
Lubello
,
C.
&
Van Loosdrecht
,
M. C. M.
2014
Physiological and kinetic characterization of a suspended cell anammox culture
.
Water Research
60
,
1
14
.
Lovley
,
D. R.
&
Phillips
,
E. J.
1998
Novel mode of microbial energy metabolism: organic carbon oxidation coupled to dissimilatory reduction of iron or manganese
.
Applied and Environmental Microbiology
54
(
6
),
1472
1480
.
Muhammad
,
A.
,
Rathnayake
,
R. M.
,
Zhang
,
L.
,
Ishii
,
S.
,
Kindaichi
,
T.
,
Satoh
,
H.
,
Toyoda
,
S.
,
Yoshida
,
N.
&
Okabe
,
S.
2016
Source identification of nitrous oxide emission pathways from a single-stage nitritation-anammox granular reactor
.
Water Research
102
,
147
157
.
Pathak
,
B. K.
,
Kazama
,
F.
,
Tanaka
,
Y.
,
Mori
,
K.
&
Sumino
,
T.
2007
Quantification of anammox populations enriched in an immobilized microbial consortium with low levels of ammonium nitrogen and at low temperature
.
Applied and Environmental Microbiology
76
(
5
),
1173
1179
.
Rathnayake
,
R. M. L. D.
,
Song
,
Y.
,
Tumendelger
,
A.
,
Oshiki
,
M.
,
Ishii
,
S.
,
Satoh
,
H.
,
Toyoda
,
S.
,
Yoshida
,
N.
&
Okabe
,
S.
2013
Source identification of nitrous oxide on autotrophic partial nitrification in a granular sludge reactor
.
Water Research
47
(
19
),
7078
7086
.
Sliekers
,
A. O.
,
Derwort
,
N.
,
Gomez
,
J. C.
,
Strous
,
M.
,
Kuenen
,
J. G.
&
Jetten
,
M. S. M.
2002
Completely autotrophic nitrogen removal over nitrite in one single reactor
.
Water Research
36
(
10
),
2475
2482
.
Vázquez-Padίn
,
J. R.
,
Pozo
,
M. J.
,
Jarpa
,
M.
,
Figueroa
,
M.
,
Franco
,
A.
,
Mosquera-Corral
,
A.
,
Campos
,
J. L.
&
Méndez
,
R.
2009
Treatment of anaerobic sludge digester effluents by the CANON process in an air pulsing SBR
.
Journal of Hazardous Materials
166
(
1
),
336
341
.
Vázquez-Padín
,
J.
,
Mosquera-Corral
,
A.
,
Campos
,
J. L.
,
Méndez
,
R.
&
Revsbech
,
N. P.
2010
Microbial community distribution and activity dynamics of granular biomass in a CANON reactor
.
Water Research
44
(
15
),
4359
4370
.
Zhang
,
X.
,
Li
,
D.
,
Liang
,
Y.
,
He
,
Y.
,
Zhang
,
Y.
&
Zhang
,
J.
2013a
Autotrophic nitrogen removal from domestic sewage in MBR-CANON system and the biodiversity of functional microbes
.
Bioresource Technology
150
,
113
120
.
Zhang
,
X.
,
Li
,
D.
,
Liang
,
Y.
,
Zhang
,
Y.
,
Fan
,
D.
&
Zhang
,
J.
2013b
Application of membrane bioreactor for completely autotrophic nitrogen removal over nitrite (CANON) process
.
Chemosphere
93
(
11
),
2832
2838
.
Zhao
,
J.
,
Huang
,
N.
,
Hu
,
B.
,
Jia
,
L.
&
Ge
,
G.
2016
Potential of nitrous oxide recovery from an aerobic/oxic/anoxic SBR process
.
Water Science and Technology
73
,
1061
1066
.